This dataset supports the study of Beral et al. (2020) including phenotyping data for two experiments conducted at INRA, Clermont-Ferrand in 2015/2016 and 2016/2017 using the Pheno3C high-throughput, field-phenotyping platform. Two genetic panels of winter wheat genotypes were indeed usedin 2015/2016 and 2016/2017 under two distinct environmental conditions. In the analysis, we have annotated E1 and E2 as respectively the first population (Panel 2016) under two contrasting water conditions; and E3 and E4 as respectively the second population (Panel 2017) under two contrasting conditions. This dataset includes one file of phenotyping data for each genetic panel. These phenotyping data are adjusted means of the following variables: SPM2: number of spikes per m², GPS: number of grains per spike, GPM2: number of grains/m², TKW: thousand kernel weight (g at 15% moisture content), GY: Grain yield (t/ha at 15% moisture content), GSM: grain size mean (mm²), GSV: grain size variance (mm^4), P5: 5th percentile of grain size distribution and P95: 95th percentile of grain size distribution. E1 (well-watered, 2016); E2 (water-deficit, 2016); E3 (well-watered, 2017); E4 (water-deficit, 2017).